Emanate Blogs

Can AI Speed Clinical Trials?

Artificial intelligence concept, with a robot hand and a human hand reaching towards each other.

These days it's hard to open your news feed or read an industry publication without seeing reference to artificial intelligence (AI) or machine learning. The technology is moving quickly, making progress towards innovations like self-driving cars and customer behavior prediction. AI offers opportunities for the life sciences industries as well, though progress has been slow. In particular, AI may be the answer to the biotech and pharma question, "How do we get life-saving therapies and devices to market faster?"

The Clinical Trial Process

The clinical trials process is slow and inefficient -- it's estimated that the process averages 7.5 years, and costs anywhere between $161M - $2B per drug or device, yet only 1 in 10 drugs that enter Phase I of a clinical trial will be approved by the FDA.

Trials can fail for many reasons, but often failure is due to problems recruiting enough patients, or an inability to retain the patients for the duration of the trial. Unintended & severe side effects, and poor data collection methods can also cause trials to fail.

According to a recent report by CBInsights, pharma companies were attempting to recruit thousands of patients for over 10,000 trials in 2018 hoping to test new life-saving cancer drugs, yet only about 5 percent of cancer patients would end up taking part in a trial.

It's evident why: For terminal illnesses in particular, like cancer, patients may not consider enrolling in a drug trial until existing forms of treatments have failed. Not only that, but not all patients diagnosed with untreatable cancer are eligible to participate, determining eligibility is difficult, and the eligibility criteria on ClinicalTrials.Gov are full of technical jargon.

His first job out of college was at a core lab, one that retained all the medical images collected during the course of a trial. Here he was responsible for the initial assessment of the images before transferring them to the physician. Basically, Kunal was an in-house data manager, ensuring the accuracy and security of the imaging data for clinical trials.

For those that are eligible, participating in a trial costs both dollars and time.

Could AI Be the Answer?

With all of the issues preventing more efficient trials, will AI be the technology that finally streamlines the process? AI is new and exciting, and being touted as the "magic bullet" for just about everything. In the clinical space, we have the Clinical Internet of Things (CIoT) for remote monitoring, machine learning for EHR processing, and AI-based cybersecurity for data protection. It appears that AI has the potential to impact every stage of a clinical trial, including choosing the best sites, tracking trial enrollment, and predicting patient enrollment curves.

Matching the right trial with the right patient is a time-consuming and challenging process for both the clinical study team and the patient."

Roughly 80% of clinical trials fail to meet enrollment timelines, and approximately one-third of Phase III clinical study terminations are due to enrollment difficulties, according to a Cognizant report on recruitment forecasts. At any given time, there could be as many as 18,000 trials recruiting patients. Patients rarely get trial recommendations from their physicians, and navigating the government website, ClinicalTrials.gov, can be difficult. These recruitment delays are costly to sponsors.

AI is poised to help both sponsor companies and patients by extracting pertinent information from medical records and comparing it to the requirements of ongoing trials. Studies matching conditions could then be recommended to patients.

One of the greatest impediments, though, is the ability, or lack thereof, to share health information easily across institutions and systems. This will continue to be a problem for AI systems.

Once a patient locates a trial, inclusion and exclusion criteria can still prevent them from enrolling in it. Tests needed to determine if a patient is a fit for a trial can require multiple visits to a clinic and take weeks. The patient must visit a participating site to see if he or she will be eligible for the trial, usually after a preliminary phone screen with a study investigator from the clinical research team.

AI could assist in this process by extracting information from medical records that could verify the inclusion and exclusion criteria. Another promising solution is patient-generated data. One company exploring this is Apple, which is gradually building a clinical study ecosystem around the iPhone and Apple Watch. By continuously monitoring patients in the comfort of their homes, the company can generate a trove of previously unavailable health data.

Adherence, Tracking, and Reporting Can Lead to Patient Dropouts

Once a patient is admitted to a trial, they need to adhere to the medication regimen. Many clinical studies still use paper diaries instead of an electronic system, and patients are expected to note when they took the study drug, what other medications were taken on those days, and any adverse reactions. Humans are fallible, prone to errors, and notoriously bad at remembering the things they've done. This is inefficient, to say the least.

Patients may drop out of any given study because of:

  • The need to record by hand, from memory, study-related activities.
  • Frequent travel to the clinic for check-ins and reporting.
  • Additional out-of-pocket costs and fees.

This is another area where technology can help.

AI Can Smooth the Reporting Process

Facial recognition technology is available that allows patients to use their phones to record videos of themselves taking medications. A healthcare companion and coach that uses AI is being developed to tailor conversations to patients, set reminders, and ask questions. The robot assistant will use a touchscreen or voice activation feature to communicate with patients.

ResearchKit and CareKit are open source frameworks available from Apple since 2015 that allow researchers and developers to create apps that monitor the daily lives of patients. Researchers at Duke University have developed an app that uses an iPhone to screen children for autism, and the mPower app uses exercises like finger tapping to study patients with Parkinson's disease.

In addition to using mobile tech to send patients reminders to take their medication, pharmaceutical companies Pfizer and Novartis have been investing in IoT and "ingestible sensors" to track drug intake. In 2017, Merck Ventures invested in Medisafe, which develops wireless pill bottles.

One potential competitor for Apple in this space is Google. Google’s Project Baseline, which aims to enroll 10,000 patients and monitor their daily lives over the course of 5 years, could ultimately benefit clinical trials and reduce recruitment bottlenecks.

Patient-generated data — like the data Project Baseline is gathering — could eliminate the need for a control group, providing the data required from the control and ultimately reducing recruitment bottlenecks. However, the project is still in its early stages.

But...AI Alone is Not the Cure

While the healthcare industry leads in AI adoption, in the actual clinical trial process AI is still in its early stages, and there’s a need for digitization that precedes the need for AI.

One of the biggest hurdles in clinical trials will be overcoming the "this is how we've always done it" mentality.

Pharma will need to understand exactly what AI can do and what its limitations will be. The first step will be to stop thinking about a futuristic state where AI eliminates all trial problems, and instead focus on achievable, short-term goals that make it easier for patients to participate in, and remain in, clinical trials.

Leave a Reply

Your email address will not be published. Required fields are marked *